In the construction of semiconductor chip assemblies, it has been found desirable to interpose encapsulating material between and/or around elements of the semiconductor assemblies in an effort to reduce and/or redistribute the strain and strain on the connections between the semiconductor chip and a supporting circuitized substrate during operation of the chip, and to seal the elements against corrosion.
Ball grid array ("BGA") packaged and chip scale packaged ("CSP") semiconductor chips and flip chip attachment solutions are connected to external circuitry through contacts on a surface of the chip. To save area on a supporting substrate, such as a printed wiring board ("PWB"), these chips are directly connected/soldered to the substrates and from there connected to external circuitry on other parts of the substrate. The chip contacts are either disposed in regular grid array patterns, substantially covering the face surface of the chip (commonly referred to as an "area array") or in elongated rows extending parallel to and adjacent each edge of the chip front surface. Many of the techniques for attachment run into problems because of the thermal expansion mismatch between the material the chip is composed of and the material the supporting circuitized substrate is made of, such as a PWB. In other words, when the chip is in operation, the chip heats up and also heats its supporting substrate thereby causing both the chip and the substrate to expand. When the heat is removed, the chip and substrate both contract. This heating and cooling process is referred to as "thermal cycling". Since the heat is being generated in the chip, the chip will heat up more quickly and will typically get hotter than its supporting substrate. The materials comprising both the chip and the substrate have inherent expansion and contraction rates, referred to as their coefficients of thermal expansion ("CTE"), which causes them to expand and contract at different rates and in different degrees when subjected to the same thermal conditions. This thermal expansion mismatch between the chips and the substrate places considerable mechanical stress and strain on the connections between the chip contacts and corresponding bond pads on the substrate.
BGA and CSP technology refers to a large range of semiconductor packages which use interconnection processes such as wirebonding, beam lead, tape automated bonding ("TAB") or the like as an intermediate connection step to interconnect the chip contacts to the exposed package terminals. This results in a device which can be tested prior to mechanical attachment to the bond pads on supporting substrate. The BGA or CSP packaged chips are then typically interconnected with their supporting substrates using standard tin-lead solder connections. In most such packaged devices, the mechanical stress/strain due to thermal cycling is almost completely placed on the solder connections between the chip and the substrate. However, solder was never intended to undergo such forces and commonly undergoes significant elastic solder deformation causing the solder to crack due to fatigue brought on by the thermal cycling. When the solder connections have smaller diameters, thermal cycling has an even more profound fatiguing affect on the solder. This has driven efforts in the packaging art to modify the solder and other elements of the packages so that they may better withstand the thermal cycling forces.
As the features of semiconductor chip packages continue to be reduced in size, as in the case of CSPs, the number of chips packed into a given area will be greater and thus the heat dissipated by the each of these chips will have a greater effect on the thermal mismatch problem. Further, the solder cracking problem is exacerbated when more than one semiconductor chip is mounted in a package, such as in a multichip module. As more chips are packaged together, more heat will be dissipated by each package which, in turn, means the interconnections between a package and its supporting substrate will encounter greater mechanical stress due to thermal cycling. Further, as more chips are integrated into multichip modules, each package requires additional interconnections thereby increasing the overall rigidity of the connection between the module and its supporting substrate.
Certain designs have reduced solder connection fatigue by redistributing the thermal cycling stress into a portion of the chip package itself. An example of such a design is shown in U.S. Pat. Nos. 5,148,265 and 5,148,266, the disclosure of which is incorporated herein by reference. One disclosed embodiment of these patents shows the use of a chip carrier in combination with a compliant layer to reduce the CTE mismatch problems. Typically, the compliant layer includes an elastomeric layer which, in the finished package, is disposed between the chip carrier and the face surface of the chip. The compliant layer provides resiliency to the individual terminals, allowing each terminal to move in relation to its electrically connected chip contact to accommodate CTE mismatch as necessary during testing, final assembly and thermal cycling of the device.
In some arrangements used heretofore, the compliant layer was formed by stenciling a thermoset resin onto the chip carrier and then curing the resin. Next, additional resin was applied to the exposed surface of the cured layer, this additional resin was partially cured, and the resulting tacky adhesive surface was used to bond the elastomeric layer to the chip and chip carrier. Once attached, the entire structure was heated and fully cured. Although this process is effective, further improvement would be desirable. The ambient gas can be occasionally entrapped when the chip carrier and die are affixed to the compliant layer. The entrapped gas can create voids and gas bubbles in the encapsulation of the surface of the die by the encapsulation material. These voids/bubbles allow moisture and other contamination to come into direct contact with the surface of the die. Accordingly, care must be taken to prevent such entrapment. This adds to the expense of the process.
In the flip-chip mounting technique, the contact bearing face surface of the chip opposes a bond pad bearing supporting substrate. Each contact on the device is joined by a solder connection to a corresponding bond pad on the supporting substrate, as by positioning solder balls on the substrate or device, juxtaposing the device with the substrate in the front-face-down orientation and momentarily reflowing the solder. The flip-chip technique yields a compact assembly, which occupies an area of the substrate no larger than the area of the chip itself. However, flip-chip assemblies suffer from significant problems when encountering thermal cycling stress because the sole thermal cycling stress bearing elements are the solder connections, as described above in relation to the BGA and CSP packages. In the case of flip chip devices, there is no package to redistribute the thermal cycling stress. Because of this, significant work has been done in the art to make the flip chip solder connections more reliable. However, to keep the chip's standoff from the substrate to a minimum, the solder connections have a typical diameter of between about five and eight thousandths of an inch ("mils"), too small to provide much real mechanical compliance. In an attempt to solve this problem, a curable liquid underfill is flowed between the chip and its attached substrate, enclosing the solder connections. The underfill is then cured into a rigid form which has a CTE that is closely matched to the solder material. The aim of the underfill is to reduce the stress caused by CTE mismatch by redistributing the stress more uniformly over the entire surface of the chip, supporting substrate and solder balls. Examples of the use of underfill materials may be found in U.S. Pat. Nos. 5,120,678, 5,194,930, 5,203,076 and 5,249,101. All of these prior art solutions are aimed at reducing the shear stress endured by the interconnections caused by thermal cycling. However, each of these solutions also encounters significant problems such as insufficient compliancy, voids and process cost. Most significant among these costs is reducing the voiding problem which occurs when the underfill flows between the chip and the substrate and traps gas therebetween. If this gas is not removed, it will typically quickly expand during a heating cycle of the chip. The force associated with the expanding gas can cause the solder connections to crack or otherwise become unreliable. Yet, presently, the underfill process involves a very costly and time-consuming process of allowing the underfill to flow very slowly between the chip and the substrate to try to avoid voids. After the underfill has flowed completely between the chip and the substrate, the assembly will then be subjected to one or more vacuuming steps in a further attempt to remove any voids in the underfill material.
Despite these and other efforts in the art, still further improvements in interconnection technology would be desirable.